Abstract

With the growth of IM technology, identity safekeeping has became an important problem. There is a constant threat of one's account being hacked or being used by an unauthorized user unknowingly. In this paper we have tried to verify the identity of the person on the basis of his writing style. Since chat is a mixture of acronyms, short forms, emoticons, symbols a lot of features can be extracted out of it, although its small size and use of regional language makes it a bit challenging. We have used per line features as in real time identifying the impostor in minimum amount of time is important. We have used the power of machine learning to learn from the history, tested with six machine learning algorithms and later combined them to improve the prediction. Maximum accuracy obtained on any user is 98% and the average accuracy is 74.58%.

Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call